They can be abused, by people who understand statistics talking to people who don’t understand statistics. This is a good reason to learn statistical methods rather than reject them.
There are levels of abuse, some blatant, some subtle. Leading questions are obvious, when you have the question asked. Publishing bias is difficult to spot, even for trained scientists looking for it.
Learning about statistical methods isn’t enough. People need to be taught how to weigh the data presented against the value of misleading them.
It’s a subsection of logical reasoning, and needs to be taught as part of an integrated whole.
A bit of healthy scientific skepticism or logical reasoning with some skills to evaluate sources of evidence and biases help with both understanding quoted stats, and liars and the ill-informed.
It’s a difficult and time consuming skill to learn and use though.
Even a small amount of statistic abuse will break blind trust in them. Once that trust is gone, some people will reject all of them, rather than try and differentiate.
Low grade abuse of statistics and related methods is rampant in low grade media.
In reality, statistics should be trusted based on source, method and importance.
A survey of preferred ice-cream flavours by an ice-cream company can be trusted easily, even if the wording and method are a bit loose. An analysis of a potentially billion dollar drug requires FAR more scrutiny, even from multiple reliable sources. Between these 2 extremes is a spectrum of trust.
Unfortunately, most people don’t do well with shades of grey. If some statistics can’t be trusted, then none can. It’s all false news (until it happens to agree with their preconceived views).
Because statistics is a relative unknown to many people. Until people have a good grounding in statistics then they often have to rely on an appeal to authority.
They can be abused, by people who understand statistics talking to people who don’t understand statistics. This is a good reason to learn statistical methods rather than reject them.
There are levels of abuse, some blatant, some subtle. Leading questions are obvious, when you have the question asked. Publishing bias is difficult to spot, even for trained scientists looking for it.
Learning about statistical methods isn’t enough. People need to be taught how to weigh the data presented against the value of misleading them.
It’s a subsection of logical reasoning, and needs to be taught as part of an integrated whole.
I think statistically (pun intended) there are more problems with people ignoring statistics or plain lying, than statistics being abused
A bit of healthy scientific skepticism or logical reasoning with some skills to evaluate sources of evidence and biases help with both understanding quoted stats, and liars and the ill-informed.
It’s a difficult and time consuming skill to learn and use though.
Even a small amount of statistic abuse will break blind trust in them. Once that trust is gone, some people will reject all of them, rather than try and differentiate.
Low grade abuse of statistics and related methods is rampant in low grade media.
But blind trust in everything else never breaks?
In reality, statistics should be trusted based on source, method and importance.
A survey of preferred ice-cream flavours by an ice-cream company can be trusted easily, even if the wording and method are a bit loose. An analysis of a potentially billion dollar drug requires FAR more scrutiny, even from multiple reliable sources. Between these 2 extremes is a spectrum of trust.
Unfortunately, most people don’t do well with shades of grey. If some statistics can’t be trusted, then none can. It’s all false news (until it happens to agree with their preconceived views).
But my point is, why does that “all or nothing” standard apply to statistics, but not to news channels, newspapers, internet articles, etc.
Because statistics is a relative unknown to many people. Until people have a good grounding in statistics then they often have to rely on an appeal to authority.